bff-entry-points

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides high-quality architectural guidance for building secure HTTP and realtime entry points, adhering to IETF and OWASP standards.
  • [SAFE]: It implements a 'deny-by-default' strategy via a prepared registrar, preventing unclassified route exposure.
  • [SAFE]: It handles session security correctly by using HttpOnly, Secure, SameSite=Strict cookies with __Host- prefixes to keep tokens server-side.
  • [EXTERNAL_DOWNLOADS]: The skill mentions standard libraries such as @hono/zod-openapi and @fastify/websocket; these are well-known, industry-standard packages appropriate for the described architecture.
  • [INDIRECT_PROMPT_INJECTION]: The instructions describe the ingestion of browser-supplied data (headers, body, parameters) but mitigate risks through mandatory strict schema validation and service-layer authorization using provider-free principals.
  • [COMMAND_EXECUTION]: No evidence of arbitrary shell commands, unsafe system calls, or dangerous subprocess execution was found in the skill or its references.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 09:52 AM
Security Audit — agent-trust-hub — bff-entry-points